Laboratory Investigation (2017) 97, 1084–1094 © 2017 USCAP, Inc All rights reserved 0023-6837/17

α--4 promotes metastasis in gastric cancer Xin Liu and Kent-Man Chu

Metastasis increases the mortality rate of gastric cancer, which is the third leading cause of cancer-associated deaths worldwide. This study aims to identify the promoting metastasis of gastric cancer (GC). A human cell motility PCR array was used to analyze a pair of tumor and non-tumor tissue samples from a patient with stage IV GC (T3N3M1). Expression of the dysregulated genes was then evaluated in GC tissue samples (n = 10) and cell lines (n =6) via qPCR. Expression of α-actinin-4 (ACTN4) was validated in a larger sample size (n = 47) by qPCR, western blot and immunohistochemistry. Knockdown of ACTN4 with specific siRNAs was performed in GC cells, and adhesion assays, transwell invasion assays and migration assays were used to evaluate the function of these cells. Expression of potential targets of ACTN4 were then evaluated by qPCR. Thirty upregulated genes (greater than twofold) were revealed by the PCR array. We focused on ACTN4 because it was upregulated in 6 out of 10 pairs of tissue samples and 5 out of 6 GC cell lines. Further study indicated that ACTN4 was upregulated in 22/32 pairs of tissue samples at stage III & IV (P = 0.0069). Knockdown of ACTN4 in GC cells showed no significant effect on cell proliferation, but significantly increased cell-matrix adhesion, as well as reduced migration and invasion of AGS, MKN7 and NCI-N87 cells. We found that NF-κB was downregulated in GC with the knockdown of ACTN4. In conclusion, this is the first study to indicate that ACTN4 is significantly upregulated in patients with metastatic GC. ACTN4 reduces cell adhesion and enhances migration and invasion of GC cells and may therefore be a novel therapeutic target for GC. Laboratory Investigation (2017) 97, 1084–1094; doi:10.1038/labinvest.2017.28; published online 5 June 2017

Gastric cancer (GC) is the third leading cause of Previous studies have indicated that numerous, complex cancer-associated deaths worldwide.1 Moreover, it has been interactions are necessary for metastasis, including predicted that the mortality rate of GC will rise from the 15th vascular endothelium growth factor receptor, matrix metal- to the 10th most common cause of death globally by 2030.2 lopeptidase (MMP), and members of the PI3K-Akt and Wnt- About 70% of the GC cases occur in eastern Asia, with the Notch signaling pathways.13–20 However, the mechanism of highest morbidity in China, Korea, and Japan. In general, the GC metastasis remains unclear; hence, the aim of this study 5-year survival rate for patients at the early stage of GC is was to identify genes promoting metastasis of GC. 490%. However, the chance for survival drops markedly to In this paper, α-actinin-4 (ACTN4) is identified as a key player o25% for advanced stage disease with metastasis.3,4 The in GC metastasis. ACTN4 is a member of the α-actinin family, preferred sites of metastatic GC include lymph nodes, liver which belongs to the superfamily . There are and peritoneum.5,6 four isoforms of α-actinin, and isoform 4 is 'non-muscle' Metastasis of cancer cells is a complex, multi-step process. α-actinin. It predominantly localizes in the cytoplasmic region,21 Generally, it involves cancer cells acquiring the capability to and has multiple roles in non-muscle cells, including cellular invade and disseminate from their original location. These protrusion, cell adhesion, and cellular motility.22,23 Functionally, cells survive in circulation and colonize in secondary organs at α-actinin is associated with bundles and tight which they form new tumors.7 Numerous genes have junctions, where it is involved in cross-linking filaments and important roles in these processes. For example, genes for connecting actin to the cell membrane. Actinin is a component chemotaxis and growth factors contribute to invasion.8,9 of the stress fiber and structural ,24 and the overexpression Genes for cell adhesion, protrusion, or movement lead to of ACTN4 contributes to metastasis and/or chemoresistance in intravasation and migration.10 Genes encoding receptors and lung cancer, ovarian cancer, and colorectal cancer.25–27 As a proteases are involved in extravasation.11,12 result, we proposed that ACTN4 is a key factor in GC metastasis.

Department of Surgery, The University of Hong Kong, Queen Mary Hospital, Pokfulam, Hong Kong Correspondence: Professor K-M Chu, MBBS, MSc, FRCS (Edin), FACS, FCSHK, FHKAM (Surgery), Department of Surgery, The University of Hong Kong, Queen Mary Hospital, Pokfulam 000, Hong Kong. E-mail: [email protected] Received 5 April 2016; revised 8 December 2016; accepted 16 January 2017

1084 Laboratory Investigation | Volume 97 September 2017 | www.laboratoryinvestigation.org α-Actinin-4 in gastric cancer X Liu and K-M Chu

MATERIALS AND METHODS SYBR Green PCR Kit (Qiagen). Specific primers for qPCR Human Tissue Samples were provided by IDT (Singapore). QPCR was performed in Forty-seven pairs of human GC and non-tumor tissue ABI ViiA7 Real-time PCR system (Applied Biosystems, samples were collected directly after surgical resection at Waltham, MA, USA). The primers sequences are listed in Queen Mary Hospital, Hong Kong. All of the tumor tissue supplementary Table 1. The expression of mRNAs was − ΔΔ samples were validated as malignant by an experienced normalized to GAPDH using the 2 Ct method. The ΔCt pathologist. All of the samples were obtained with the was calculated by subtracting the Ct values of GAPDH from participants’ informed consent and none of the patients the Ct values of targets. The ΔΔCt was then calculated by received preoperative treatment. All samples were immedi- subtracting ΔCt of non-tumor sample from ΔCt of the paired ately frozen in liquid nitrogen and stored at − 80 °C. tumor sample. Fold changes in the targets were calculated by − ΔΔ the equation 2 Ct. Cell Lines and Cell Culture Human GC cell lines including AGS, NCI-N87 (ATCC, Western Blots Rockville, MD, USA), MKN7, MKN45 (Riken, Japan) and Cells were harvested and lysed by RIPA Buffer (Sigma BGC823, PAM82 (given by Beijing Institute of Cancer Chemical, St Louis, MD, USA). Samples containing equal Research, China) were used in this study. Cells were cultivated amounts of protein were separated by SDS-PAGE and in RPMI 1640 medium (Gibco BRL, Gaithersburg, MD, USA) electroblotted onto Immobilon-P Transfer Membrane supplemented with 10% heat-inactivated fetal bovine serum (Applied Biosystems). The membrane was blocked with 5% (FBS) (Gibco BRL, Grand Island, NY, USA). All cells were non-fat milk, followed by incubation with antibodies specific incubated at 37 °C in a humidified incubator which contains for anti-ACTN4 (1:1000, Abcam, Cambridge, UK) and anti- 5% CO2. β-actin (1:10 000, Cell Signaling Technology, Beverly, MA, USA), respectively. Blots were then incubated with goat anti- Rna Extraction in Tissue Samples and Cell Lines rabbit or anti-mouse secondary antibody conjugated to About 20 mg of each tissue sample or 106 cells of each cell line horseradish peroxidase (Amersham Pharmacia, Cleveland, were used for RNA extraction via RNeasy Mini Kit (Qiagen, OH, USA) accordingly. The signals were captured by FUJI Hilden, Germany) according to the manufacturer’s instruc- Medical X-Ray Film (Fuji, Japan) and developed by the FUJI tions. The concentrations of all RNA samples were quantified system. by NanoDrop 1000 (Nanodrop, Wilmington, DE, USA). From each sample, 500 ng of total RNA were utilized for Immunohistochemistry reverse transcription (RT). Tissue sections were obtained from tumor and non-tumor segment of patients with GC. Consecutive paraffin-embedded Rt2 Profilertm PCR Array for Human Cell Motility tissue sections of 5 μm were dewaxed and dehydrated. The A pair of tumor and non-tumor tissue samples from a patient slides were incubated at 50 °C overnight, then soaked in with GC (stage IV with metastasis, T3N3M1) was used for the xylene for 10 min three times, and immersed in 100% ethanol PCR array. Total RNA was extracted as prescribed above and for 5 min, 95% ethanol for 5 min, and 80 and 70% ethanol for reverse-transcribed using RT2 First Strand Kit (Qiagen). The 1 min. After rehydration for 5 min, the slides were boiled in cDNAs was used for the PCR array using RT2 SYBR Green citrate buffer (pH 6.1, Target Retrieval Solution; Sigma, CA, qPCR Mastermixes (Qiagen). This profiler is a pathway- USA) for 25 min, rinsed with 1 × PBS for 5 min, and treated focused analysis. It contains 84 genes with 3% hydrogen peroxide for 10 min. After that, slides were involved in cell movement, including cell adhesion, cellular placed flat on the bench top and sections were covered in a projection, chemotaxis, proteases, and their inhibitors. By solution of rabbit antibody to ACTN4 (1:100 dilution; comparing gene profiles between tumor and non-tumor Abcam) at room temperature for 30 min and then covered tissue samples, different gene expression profiles were with a solution of goat anti-rabbit (1:500) (Dako) for 30 min, established. Upregulation of the genes (greater than twofold) and finally in a solution of streptavidin–horseradish perox- in tumor samples compared with non-tumor samples after idase (LSAB2 System; Dako) for 30 min. Color was with the normalization were chosen for further study. Substrate—Chromogen Solution (LSAB2 System; Dako) for 5 min, and counter-stained with Mayer hematoxylin (Merck, RT Quantitative PCR (qPCR) Darmstadt, Germany) for 1 min. Finally, the slides were SYBR green RT quantitative PCR (RT qPCR) assay was used dehydrated in 50, 75, and 100% ethanol for 5 min each, and for mRNA quantification in tissue samples and cells. A total sealed with mounting medium (CytosealTM 60 Thermo of 500 ng of RNA from each tissue sample or cell line Scientific, Waltham, MA, USA). containing total RNA were polyadenylated by poly (A) polymerase and reverse-transcribed to cDNA using Omnis- SiRNAs and Transfection cript RT Kit (Qiagen) according to the manufacturer’s Small interfering RNAs (TriFECTa RNAi Kit) targeting instructions. Real-time PCR was performed using miScript ACTN4 were purchased from Integrated DNA Technologies,

www.laboratoryinvestigation.org | Laboratory Investigation | Volume 97 September 2017 1085 α-Actinin-4 in gastric cancer X Liu and K-M Chu

(Coralville, USA) (Supplementary Table 2). A total of 5 × 104 according to the manufacturer’s instructions. Transient cells were seeded into a 12-well plate a day in advance of transfected cells were evaluated for expression levels and transfection and transfected with 20 nM siRNA or scrambled functional assays. control using Hiperfect Transfection Reagent (Qiagen), Adhesion Assay Cells with transfection of siRNAs or scrambled control were Table 1 Dysregulated genes in human cell motility profiler incubated for 24 h were harvested and employed for cell adhesion assays (CBA-052, Cell Biolabs, CA, USA). Forty- Symbol Full name GenBank T/N eight-well plates were coated with collagen I or BSA, and 50 000 cells in 150 μl of serum-free RPMI 1640 medium were Cell adhesion seeded into each well. The plates were incubated at 37 °C for ACTN1 Actinin, alpha 1 NM_001102 3.54 1 h, followed by aspiration of the medium and washing by BCAR1 Breast cancer anti-estrogen resistance 1 NM_014567 8.07 PBS to remove the non-adhesive cells. The wells were stained ENAH Enabled homolog (Drosophila) NM_1008493 15.45 with cell stain solution for 10 min. The adhesive stained cells RASA1 RAS p21 protein activator 1 NM_002890 2.06 were photo-captured. the stained cells were extracted by the VASP Vasodilator-stimulated phosphoprotein NM_003370 3.13 extraction solution. Finally, extracted samples were trans- VCL NM_003373 4.61 ferred to a 96-well plate and measured the OD 570 nm by a microplate photometer.

Cell projection Migration Assay and Invasion Assay ACTN4 Actinin, alpha 4 NM_004924 3.49 Cells transfected with siRNAs or scrambled controls (24- h CTTN Cortactin NM_005231 5.92 incubation) were harvested and suspended in RPMI 1640 DIAPH1 Diaphanous homolog 1 NM_005219 4.74 medium. A transwell culture insert with 8 μm pore size RDX Radixin NM_002906 4.42 membrane for 24-well plate (Cat no. 353097, Falcon, NY, RHOB Ras homolog gene family, member B NM_004040 3.15 USA) was used to analyze the migration activity. A total of μ RHOC Ras homolog gene family, member C NM_175744 3.44 30 000 suspension of cells in 300 l of serum-free RPMI 1640 μ SVIL Supervillin NM_003174 6.57 medium were seeded into the upper inserted chamber, 700 l of RPMI containing 10% FBS were added to the well, and the SRC V-src sarcoma viral oncogene homolog NM_005417 4.79 plate was incubated at 37 °C for 24 h. After incubation, the inner wall of the chamber was wiped with swabs to remove Chemotaxis non-migrating cells. The outer wall of the chamber was gently EGFR Epidermal growth factor receptor NM_005228 4.68 rinsed with PBS and stained with Crystal Violet (Sigma- ITGB3 Integrin, beta 3 (antigen CD61) NM_000212 10.4 Aldrich, St Louis, MO) for 10 min. Finally, the membrane was LIMK1 LIM domain kinase 1 NM_002314 3.76 rinsed and allowed to air-dry. The stained membrane was MYH9 , heavy chain 9, non-muscle NM_002473 3.76 photographed and the number of migrated cells was counted. Invasion assays were performed with a modified migration PAK1 P21 protein (Cdc42/Rac)-activated kinase 1 NM_002576 4.78 assay, described above, using Matrigel-coated transwell PAK4 P21 protein (Cdc42/Rac)-activated kinase 4 NM_005884 3.18 chambers with 8 μm pore size membranes in 24-well plates PLCG1 Phospholipase C, gamma 1 NM_002660 4.17 (Corning, NY, USA). A total of 30 000 cells in 300 μl serum- PRKCA Protein kinase C, alpha NM_002737 4.98 free RPMI 1640 were seeded onto the Matrigel over the upper TGFβ1 Transforming growth factor, beta 1 NM_000660 3.12 chamber, followed by the addition of 700 μl of RPMI VEGFA Vascular endothelial growth factor A NM_003376 16.38 containing 10% FBS at the bottom insert-well. After incubation of the cells at 37 °C for 48 h, the following steps were performed as the same of migration assay. Proteases Akt1 V-akt murine thymoma viral oncogene NM_005163 3.17 Statistical Analysis homolog 1 Statistical analysis was carried out using Statistical Package for CAPN2 Calpain 2, (m/II) large subunit NM_001748 4.49 Social Sciences (SPSS) 16.0 for Windows (SPSS, Chicago, IL, MMP14 Matrix metallopeptidase 14 NM_004995 3.47 USA). Student’s t-test was used to analyze the results MMP2 Matrix metallopeptidase 2 NM_004530 5.40 expressed as mean ± s.d. Wilcoxon sign rank test was used MMP9 Matrix metallopeptidase 9 NM_004994 3.68 to analyze ACTN4 in upregulation of GC. All P-values are PLAUR Plasminogen activator, urokinase receptor NM_002659 4.14 two-sided and a value of P ≤ 0.05 is considered statistically significant.

1086 Laboratory Investigation | Volume 97 September 2017 | www.laboratoryinvestigation.org α-Actinin-4 in gastric cancer X Liu and K-M Chu

RESULTS Figure 1a represent the ratios of expression of genes in tumor Human Cell Motility in GC Tissue Samples tissue samples to the paired non-tumor tissue samples. Values RT2 Profiler PCR Array for Human Cell Motility was above 1 represent that the gene’s expression in tumor tissue performed in a pair of tumor and non-tumor tissue samples samples is higher than that in the paired non-tumor tissue from the same patient with GC (stage IV with metastasis, samples, and the number indicates upregulation of each gene T3N3M1). Expression of 84 genes were evaluated from in these 10 samples. The qPCR result indicates that ACTN4, this PCR array. Gene expression was normalized to internal BCAR1, CAPN2, DIAPH1, PLCG1, and RASA1 are upregu- controls of GAPDH and β-actin. Upon comparison of lated in 6/10, 3/10, 4/10, 4/10, 6/10, and 4/10 pairs of gastric expression patterns in the tissue samples, a panel of tissue samples (Figure 1a). dysregulated genes was revealed. An average of a twofold In addition to tissue samples, gene expression was also difference in expression was chosen as the cutoff point. Thirty evaluated in GC cell lines. The cell lines used are as follows: upregulated genes were found in this PCR array (Table 1). MKN7, which was established from a GC metastasizing to Among these genes, several of them are consistent with lymph nodes; and MKN45 and NCI-N87, established from previous studies, such as Ras associated genes-RhoB and GC metastasizing to liver. These sites are favored for RhoC, EGFR, TGF-β, Akt, and MMP9. Remarkably, we also metastasis of GC. QPCR results show that ACTN4, DIAPH1, found a number of the upregulated genes have never been PLCG1, and RASA1 are upregulated in 5/6, 2/6, 4/6, and 4/6 reported as associated with GC before, including ACTN4, GC cell lines compared with commercial normal stomach BCAR1, CAPN2, DIAPH1, PLCG1, and RASA1.These results tissue (CR561840, Origene, MD, USA) (Figure 1b). As suggest that there are a diverse panel of genes that may play ACTN4 is upregulated in 6 out of 10 pairs of tissue samples significant roles in GC metastasis. and 5 out of 6 cancer cell lines, we focused on ACTN4 for further investigation. Expression of Genes in Tissue Samples and Cell Lines Expression of the selected genes was evaluated in 10 pairs of Actn4 is Upregulated in GC Tissue Samples and Cell gastric tumor/non-tumor tissue samples and six GC cell lines Lines via qPCR. The tissue samples were collected from patients in To validate upregulation of ACTN4 in GC, we evaluated the stage III or stage IV with metastasis. The values (of y axis) in expression of ACTN4 in 47 pairs of tissue samples. The

Figure 1 Expression of upregulated genes in human gastric cancer tissue samples and cell lines. Expression of selected genes was evaluated in 10 pairs of stage III or stage IV gastric cancer tissue samples and six gastric cancer cell lines via qPCR. (a) QPCR indicates that ACTN4, BCAR1, CAPN2, DIAPH1, PLCG1, and RASA1 are upregulated in 6/10, 3/10, 4/10, 4/10, 6/10, and 4/10 pairs of tissue samples. (b). ACTN4, DIAPH1, PLCG1, and RASA1 are upregulated in 5/6, 2/6, 4/6, and 4/6 gastric cancer cell lines comparing with commercial normal stomach tissue.

www.laboratoryinvestigation.org | Laboratory Investigation | Volume 97 September 2017 1087 α-Actinin-4 in gastric cancer X Liu and K-M Chu

Table 2 Clinical characteristics of the subjects in this study indicated). ACTN4 stain was also found in several adjacent non-tumor tissue samples (Supplementary Figure 2). This Characteristics GC patients (N = 47) result is consistent with the qRT-PCR analysis (Figure 2).

Age (years) Knockdown of Actn4 in GC Cells – Mean (range) 66 (36 89) As ACTN4 is overexpressed in GC, knockdown of ACTN4 Median 67 was performed in GC cells with specific siRNAs (TriFECTa RNAi Kit, IDT). The red fluorescent signal indicates that the Gender percentage of transfection of siRNAs in GC cells reaches over Male 30 90% in AGS, MKN7, and NCI-N87 (Figure 4a). At both the mRNA and protein levels, expression of ACTN4 is signifi- Female 17 cantly decreased in these cell lines 72 h after knockdown (Figures 4b and c). As qPCR is a very sensitive analytic Lauren classification method, it appears that there is a difference between the Intestinal 23 expression levels of mRNA and protein with knockdown of Diffuse 15 ACTN4 in AGS and NCI-N87 cells. But both sets of data Intermediate 2 indicate that ACTN4 is significantly downregulated by specific siRNAs in these cells. The cells with knockdown of ACTN4 Not Specified 7 were analyzed in functional assays.

Differentiation Downregulation of Actn4 Enhances GC Cell Adhesion Well 0 Cell adhesion assay results with AGS, MKN7, and NCI-N87 Moderate 16 cells are shown in Figure 5a. Cells did not adhere to plates Poor 27 treated with BSA (data not shown). Relative OD values are indicated in Figure 5b. The result reveals that knockdown of Not Specified 4 ACTN4 enhances GC cell-matrix adhesion by 40–100% (*Po0.05, **Po0.01). From these data, we conclude that the TNM stages overexpression of ACTN4 inhibits cell adhesion and con- I8tributes to cell migration. II 7 III 16 Downregulation of Actn4 Suppresses Migration and Invasion of GC Cells IV 16 To further validate that ACTN4 enhances GC metastasis, we employed a transwell migration assay with AGS, MKN7, and NCI-N87 cells that overexpress ACTN4. The cells that had clinical characteristics of these patients are listed in Table 2. migrated were stained and were counted by microscopy. QPCR results indicate that ACTN4 is upregulated in 22 out of Representative images of stained cells with scrambled control 32 GC tissue samples in advanced stage III and IV with or siRNAs for ACTN4 are shown in Figure 6a. Relative cell metastasis (*P = 0.0069), whereas it is upregulated in 4 out of numbers are indicated in Figure 6b. The migration assay 15 samples in early stage I and II (P = 0.4212) (Figure 2a). reveals that migrated GC cells are significantly decreased by – Po Overexpression of ACTN4 in GC cell lines was also validated 40 60% with reduction of ACTN4 (** 0.01). at protein level, except for in BGC823 cells (Figure 2b). This For invasion assay, the transwell chambers were coated with Matrigel to mimic the basement membrane in the tumor may be attributed to the post-transcriptional regulation of microenvironment. The invasion assay also reveals that the ACTN4 in this cell line. As MKN7, MKN45, and NCI-N87 invasiveness of GC cells are significantly decreased by are established from GC metastasizing to favorite sites lymph knockdown of ACTN4 in AGS and MKN7 cells by ~ 40– node and liver, the result suggests that ACTN4 is significantly 50% (Figures 7a and b), **Po0.01. The above data indicate overexpressed in metastatic GC. that the upregulation of ACTN4 promotes GC cell migration The above data suggest that ACTN4 is overexpressed in and invasion. advanced stage of GC tissue samples. Immunohistochemical analysis of ACTN4 expression in these samples was also Nf-kb is Decreased in GC Cells and Tissue Samples with employed to verify that the high expression of ACTN4 is Downregulation of Actn4 intrinsic to cancer cells. ACTN4 expression in FFPE samples To explore the mechanism of ACTN4 promotion of was detected by immunohistochemistry, which shows that metastasis of GC, we evaluated the expression of its potential ACTN4 is expressed in cancer cells (Figure 3, arrows targets in GC cells and tissue samples. We found that the

1088 Laboratory Investigation | Volume 97 September 2017 | www.laboratoryinvestigation.org α-Actinin-4 in gastric cancer X Liu and K-M Chu

Figure 2 ACTN4 is overexpressed in human gastric cancer tissue samples and cell lines. (a) QPCR indicates that ACTN4 is upregulated in 22 out of 32 pairs of gastric cancer tissue samples in advanced stage III and IV with metastasis (**P = 0.0069), whereas it is upregulated in 4 out of 15 pairs of samples in early stage I and II (P = 0.4212). (b) Overexpression of ACTN4 in gastric cancer cell lines is validated in protein level.

Figure 3 Immunohistochemical analysis of ACTN4 expression in tissue samples of gastric cancer patients. Representative images of the ACTN4 protein immunoreactivity of gastric cancer tissue samples, under a light microscope (×400 magnification). Tissue samples were stained with ACTN4-specific antibody as indicated. Cytoplasmic staining with ACTN4 antibody was detected (ACTN4-brown; nuclear DNA-hematoxylin & eosin). Arrows indicate high expression of ACTN4.

www.laboratoryinvestigation.org | Laboratory Investigation | Volume 97 September 2017 1089 α-Actinin-4 in gastric cancer X Liu and K-M Chu

Figure 4 Knockdown of ACTN4 in gastric cancer cells. Knockdown of ACTN4 was performed in gastric cancer cells with specific siRNAs. (a) Fluorescent signal indicates that the percentage of transfection reached more than 90% in AGS, MKN7, and NCI-N87 cells. (b and c) Both mRNA and protein level indicate that expression of ACTN4 is significantly decreased in three cell lines after knockdown of 72 h.

subunit NF-κB p50 is significantly downregulated in AGS and challenge in cancer management.28 For GC, the 5-year NCI-N87 cells, and p65 is downregulated in AGS and MKN7 survival rate is o25% for patients with advanced stage with cells, with knockdown of ACTN4 (Figure 8a), *Po0.05, metastasis, compared with a 490% survival rate for patients **Po0.01. Further analyses of expression of ACTN4 and in early stages without metastasis. Hence, there is an urgent NF-κB p50 in GC tissue samples indicate that the down- need to elucidate the mechanism of metastasis in order to regulation of ACTN4 is correlated with decrease of NF-κB overcome chemoresistance and to provide better prognosis p50, especially in stage III and IV samples (Figure 8b). for patients with GC. Previous studies have demonstrated that NF-κB has a Cancer metastasis is a complicated, multi-step process. significant role in development and progression of GC, and Acquisition of metastatic characteristics by cancer cells is associated with prognosis. These data suggest that ACTN4 induces chemoresistance and further leads to the mortality may regulate NF-κB to promote metastasis of GC. rate of this disease. To acquire metastatic capability, GC cells interact with tumor microenvironment. The transformed cells DISCUSSION receive signals to invade the basement membrane, escape Metastasis is the most malignant characteristic of cancer. It from their original location, move and survive throughout accounts for cancer mortality and remains as a major circulation, extravasate, and finally colonize at distant sites.29

1090 Laboratory Investigation | Volume 97 September 2017 | www.laboratoryinvestigation.org α-Actinin-4 in gastric cancer X Liu and K-M Chu

Figure 5 Inhibition of ACTN4 enhances cell adhesion in gastric cancer cells. (a) Representative images of stained adhesive AGS, MKN7, and NCI-N87 cells to collagen I are shown. (b) Stained cells were extracted and measured by microplate photometer. Relative OD values are indicated. The results are shown in mean ± s.d. *Po0.05, **Po0.01.

Decreased cell–cell or cell–matrix adhesion and increased cancer.26,33,34 Moreover, amplification of 19q13, migratory capability have critical roles in metastasis of GC. where the ACTN4 gene located, is also reported in GC.35,36 We proposed that genes associated with cell adhesion and This finding suggests that overexpression of ACTN4 in GC motility played significant roles in the process of metastasis. may also be attributed to chromosome amplification. In this study, we found a number of genes differentially It has been reported that ACTN4 suppresses expressed in tumor tissue samples compared with paired non- maturation in colorectal cancer.37 Recruitment of ZYX in tumor tissue samples in patients with GC. It is the first study focal adhesion is crucial in adhesion maturation and in to demonstrate that ACTN4 has a significant role in maintaining the stability of cell adhesion.38–40 QPCR indicates metastasis of GC. Upregulation of ACTN4 is also observed that mRNA of ZYX is not regulated by knockdown of ACTN4 in GC cell lines from favorite metastatic sites of lymph nodes (data not shown). These results suggest that ACTN4 may (MKN7) and liver (MKN45 and NCI-N87). Cancer metas- regulate ZYX post-transcriptionally. Whether ACTN4 sup- tasis to these sites is associated with poor prognosis.30–32 In presses focal adhesion through ZYX in GC should be further addition, ACTN4 does not have significant effect on GC cell investigated. Another study indicates that increase of proliferation in MTT assay (Supplementary Figure 1), which β- and actinin-4 is observed in colorectal cancer cells may be attributed to ACTN4 is not upregulated in early stage (CRC) with downregulation of E-cadherin. Inhibition of of GC. This finding suggests that overexpression of ACTN4 in E-cadherin induces β-catenin and actinin-4 proteins into the advanced stages of GC is crucial to metastasis. The data in this membrane protrusions of CRC cells.41 It has also been shown study validate the hypothesis that ACTN4 suppresses cell that reduction of E-cadherin is associated with high adhesion, promotes cell invasion and increases cell migration expression of actinin-4 in ovarian cancer patients.42 These of GC and contributes to metastasis of GC. results suggest that upregulation of actinin-4 may also be We also explored the potential mechanism of overexpres- attributed to downregulation of E-cadherin. As loss of sion of ACTN4 and its functions in GC development. E-cadherin is seen in diffuse-type GC, overexpression of Previous studies indicate that the copy number of ACTN4 ACTN4 in diffuse GC may be associated with downregulation increases in lung cancer, pancreatic cancer, and ovarian of E-cadherin. Other studies reveal that ACTN4 is a

www.laboratoryinvestigation.org | Laboratory Investigation | Volume 97 September 2017 1091 α-Actinin-4 in gastric cancer X Liu and K-M Chu

Figure 6 Inhibition of ACTN4 suppresses migration of gastric cancer cells. Transwell migration assay of AGS, MKN7 and NCI-N87 cells. (a) The representative images of stained migrated cells with scrambled control or siRNAs of ACTN4 are shown. (b) The relative numbers of migrated cells are indicated. The ratios are shown in mean ± s.d. **Po0.01.

Figure 7 Inhibition of ACTN4 suppresses invasion of gastric cancer cells. Transwell invasion assay of AGS and MKN7 cells. (a) The representative images of stained invasive cells with scrambled control or siRNAs of ACTN4 are shown. (b) The relative numbers of invasive cells are indicated. The ratios are shown in mean ± s.d. **Po0.01.

1092 Laboratory Investigation | Volume 97 September 2017 | www.laboratoryinvestigation.org α-Actinin-4 in gastric cancer X Liu and K-M Chu

Figure 8 Expression of NF-κB in gastric cancer cells and clinical tissue samples with downregulation of ACTN4. (a) Expression of NF-κB in AGS, MKN7, and NCI-N87 cells with knockdown of ACTN4 via qPCR is indicated. *Po0.05, **Po0.01. (b) Expression of ACTN4 and NF-κB p50 in clinical tissue samples are shown. The values indicate the relative expression of genes in tumor tissue samples with respective to their paired adjacent non-tumor tissue samples.

downstream target of PI3K/Akt pathway, ACTN4 regulates metastases. ACTN4 may therefore be a novel therapeutic RhoA and promotes expression of NF-κB.43–46 It is known target for metastatic GC. that PI3K/Akt pathway is crucial in the development of drug resistance.47–49 Drug resistance is a serious consequence of Supplementary Information accompanies the paper on the Laboratory Investigation website (http://www.laboratoryinvestigation.org) cancer cell metastasis. As a result, targeting ACTN4 not only inhibits metastasis, but also suppresses chemoresistance. In addition, we find that p50 and p65, subunits of NF-κB, are ACKNOWLEDGMENTS This project was supported by the Small Project Funding for basic research downregulated in GC cells with knockdown of ACTN4. This from the University of Hong Kong (201409176102). Sincere thanks are sent to finding suggests that ACTN4 may regulate NF-κBinGC. Dr Tammy Pang for technical support of sample preparation and western blot Previous studies indicate that ACTN4 functions as a in this study. Sincere thanks are also sent to Mr Leo Wu for language editing transcriptional co-activator of subunits p50 and p65 of NF- of this manuscript. κB in HEK293 and A431 cells. ACTN4 has an important role in nuclear translocation and the functions of NF-κB.46,50 Our DISCLOSURE/CONFLICT OF INTEREST study suggests that downregulation of ACTN4 contributes to The authors declare no conflict of interests for this study. a decrease in NF-κB p50 in metastatic GC cells and clinical tissue samples, which suggests that downregulation of NF-κB 1. Torre LA, Bray F, Siegel RL, et al. Global cancer statistics, 2012. CA Cancer J Clin 2015;65:87–108. may partly be attributed to knockdown of ACTN4 in GC. As 2. Mathers CD, Loncar D. Projections of global mortality and burden of NF-κB has critical roles in cancer development including disease from 2002 to 2030. PLoS Med 2006;3:e442. 51–53 κ 3. Leung WK, Wu MS, Kakugawa Y, et al. Screening for gastric cancer in metastasis, decrease of NF- B via inhibition of ACTN4 Asia: current evidence and practice. Lancet Oncol 2008;9:279–287. may abrogate metastasis of GC. It is believed that ACTN4 4. Dinis-Ribeiro M, Areia M, de Vries AC, et al. Management of interacts with numerous genes and components to regulate precancerous conditions and lesions in the stomach (MAPS): guideline from the European Society of Gastrointestinal Endoscopy (ESGE), signaling pathways leading to metastasis of GC. European Helicobacter Study Group (EHSG), European Society of In conclusion, this is the first study to reveal that ACTN4 Pathology (ESP), and the Sociedade Portuguesa de Endoscopia – contributes to metastasis of GC by suppressing cell adhesion, Digestiva (SPED). Endoscopy 2012;44:74 94. 5. Kwee RM, Kwee TC. Modern imaging techniques for preoperative enhancing cell invasion, and migration. Elucidation of the detection of distant metastases in gastric cancer. World J Gastroenterol mechanism of metastasis may help prevent metastasis in 2015;21:10502–10509. 6. Kim KH, Lee KW, Baek SK, et al. Survival benefit of gastrectomy +/- patients who are diagnosed with early cancer lesions, as well metastasectomy in patients with metastatic gastric cancer receiving as to lead to effective therapies for patients with established chemotherapy. Gastric Cancer 2011;14:130–138.

www.laboratoryinvestigation.org | Laboratory Investigation | Volume 97 September 2017 1093 α-Actinin-4 in gastric cancer X Liu and K-M Chu

7. Valastyan S, Weinberg RA. Tumor metastasis: molecular insights and 33. Noro R, Honda K, Tsuta K, et al. Distinct outcome of stage I lung evolving paradigms. Cell 2011;147:275–292. adenocarcinoma with ACTN4 cell motility gene amplification. Ann 8. Zlotnik A, Burkhardt AM, Homey B. Homeostatic chemokine receptors Oncol 2013;24:2594–2600. and organ-specific metastasis. Nat Rev Immunol 2011;11:597–606. 34. Watanabe T, Ueno H, Watabe Y, et al. ACTN4 copy number increase as 9. Wakefield LM, Hill CS. Beyond TGFbeta: roles of other TGFbeta a predictive biomarker for chemoradiotherapy of locally advanced superfamily members in cancer. Nat Rev Cancer 2013;13:328–341. pancreatic cancer. Br J Cancer 2015;112:704–713. 10. Ebos JM, Kerbel RS. Antiangiogenic therapy: impact on invasion, disease 35. Varis A, van Rees B, Weterman M, et al. DNA copy number changes in progression, and metastasis. Nat Rev Clin Oncol 2011;8:210–221. young gastric cancer patients with special reference to 11. Reymond N, d'Agua BB, Ridley AJ. Crossing the endothelial barrier . Br J Cancer 2003;88:1914–1919. during metastasis. Nat Rev Cancer 2013;13:858–870. 36. van Dekken H, Alers JC, Riegman PH, et al. Molecular cytogenetic 12. Tauro M, McGuire J, Lynch CC. New approaches to selectively target evaluation of gastric cardia adenocarcinoma and precursor lesions. Am cancer-associated matrix metalloproteinase activity. Cancer Metastasis J Pathol 2001;158:1961–1967. Rev 2014;33:1043–1057. 37. Fukumoto M, Kurisu S, Yamada T, et al. alpha-Actinin-4 enhances 13. Lim J, Ryu JH, Kim EJ, et al. Inhibition of vascular endothelial growth colorectal cancer cell invasion by suppressing focal adhesion factor receptor 3 reduces migration of gastric cancer cells. Cancer maturation. PLoS One 2015;10:e0120616. Invest 2015;33:398–404. 38. Hoffman LM, Jensen CC, Chaturvedi A, et al. Stretch-induced actin 14. Park, do J, Thomas NJ, Yoon C, et al. Vascular endothelial growth factor remodeling requires targeting of to stress fibers and recruitment a inhibition in gastric cancer. Gastric Cancer 2015;18:33–42. of actin regulators. Mol Biol Cell 2012;23:1846–1859. 15. Li W, Li S, Deng L, et al. Decreased MT1-MMP in gastric cancer 39. Gregory Call S, Brereton D, Bullard JT, et al. A zyxin-nectin interaction suppressed cell migration and invasion via regulating MMPs and EMT. facilitates zyxin localization to cell-cell adhesions. Biochem Biophys Res Tumour Biol 2015;36:6883–6889. Commun 2011;415:485–489. 16. Chen SZ, Yao HQ, Zhu SZ, et al. Expression levels of matrix 40. Sun N, Huiatt TW, Paulin D, et al. interacts with the LIM metalloproteinase-9 in human gastric carcinoma. Oncol Lett 2015;9: domain protein zyxin and is essential for cell adhesion and migration. 915–919. Exp Cell Res 2010;316:491–505. 17. Hao NB, Tang B, Wang GZ, et al. Hepatocyte growth factor (HGF) 41. Hayashida Y, Honda K, Idogawa M, et al. E-cadherin regulates the upregulates heparanase expression via the PI3K/Akt/NF-kappaB signaling association between beta-catenin and actinin-4. Cancer Res 2005;65: pathway for gastric cancer metastasis. Cancer Lett 2015;361:57–66. 8836–8845. 18. Liu J, Zhang Y, Xu R, et al. PI3K/Akt-dependent phosphorylation of 42. Yamamoto S, Tsuda H, Honda K, et al. Actinin-4 expression in ovarian GSK3beta and activation of RhoA regulate Wnt5a-induced gastric cancer: a novel prognostic indicator independent of clinical stage and cancer cell migration. Cell Signal 2013;25:447–456. histological type. Mod Pathol 2007;20:1278–1285. 19. Huang J, Xiao D, Li G, et al. EphA2 promotes epithelial-mesenchymal 43. Lu Z, Zhang Y, Yan X, et al. Estrogen stimulates the invasion of ovarian transition through the Wnt/beta-catenin pathway in gastric cancer cancer cells via activation of the PI3K/AKT pathway and regulation of cells. Oncogene 2014;33:2737–2747. its downstream targets Ecadherin and alphaactinin4. Mol Med Rep 20. Hsu KW, Hsieh RH, Huang KH, et al. Activation of the Notch1/STAT3/ 2014;10:2433–2440. Twist signaling axis promotes gastric cancer progression. Carcino- 44. Quick Q, Skalli O. Alpha-actinin 1 and alpha-actinin 4: contrasting roles genesis 2012;33:1459–1467. in the survival, motility, and RhoA signaling of astrocytoma cells. Exp 21. Murphy AC, Young PW. The actinin family of actin cross-linking Cell Res 2010;316:1137–1147. proteins - a genetic perspective. Cell Biosci 2015;5:49. 45. Zhao X, Hsu KS, Lim JH, et al. alpha-Actinin 4 potentiates nuclear factor 22. Ehrlicher AJ, Krishnan R, Guo M, et al. Alpha-actinin binding kinetics kappa-light-chain-enhancer of activated B-cell (NF-kappaB) activity in modulate cellular dynamics and force generation. Proc Natl Acad Sci podocytes independent of its cytoplasmic actin binding function. J Biol USA 2015;112:6619–6624. Chem 2015;290:338–349. 23. Hamill KJ, Hopkinson SB, Skalli O, et al. Actinin-4 in keratinocytes 46. Aksenova V, Turoverova L, Khotin M, et al. Actin-binding protein regulates motility via an effect on lamellipodia stability and matrix alpha-actinin 4 (ACTN4) is a transcriptional co-activator of RelA/p65 adhesions. Faseb J 2013;27:546–556. sub-unit of NF-kB. Oncotarget 2013;4:362–372. 24. Feng Y, Ngu H, Alford SK, et al. alpha-actinin1 and 4 tyrosine 47. Sukawa Y, Yamamoto H, Nosho K, et al. HER2 expression and PI3K-Akt phosphorylation is critical for stress fiber establishment, maintenance pathway alterations in gastric cancer. Digestion 2014;89:12–17. and focal adhesion maturation. Exp Cell Res 2013;319:1124–1135. 48. Park J, Ko YS, Yoon J, et al. The forkhead transcription factor FOXO1 25. Gao Y, Li G, Sun L, et al. ACTN4 and the pathways associated with cell mediates cisplatin resistance in gastric cancer cells by activating motility and adhesion contribute to the process of lung cancer phosphoinositide 3-kinase/Akt pathway. Gastric Cancer 2014;17: metastasis to the brain. BMC Cancer 2015;15:277. 423–430. 26. Yamamoto S, Tsuda H, Honda K, et al. Actinin-4 gene amplification in 49. Yu HG, Ai YW, Yu LL, et al. Phosphoinositide 3-kinase/Akt pathway ovarian cancer: a candidate oncogene associated with poor patient plays an important role in chemoresistance of gastric cancer cells prognosis and tumor chemoresistance. Mod Pathol 2009;22:499–507. against etoposide and doxorubicin induced cell death. Int J Cancer 27. Honda K, Yamada T, Hayashida Y, et al. Actinin-4 increases cell motility 2008;122:433–443. and promotes lymph node metastasis of colorectal cancer. Gastro- 50. Babakov VN, Petukhova OA, Turoverova LV, et al. RelA/NF-kappaB enterology 2005;128:51–62. transcription factor associates with alpha-actinin-4. Exp Cell Res 28. Sethi N, Kang Y. Unravelling the complexity of metastasis - molecular 2008;314:1030–1038. understanding and targeted therapies. Nat Rev Cancer 2011;11:735–748. 51. Li J, Deng Z, Wang Z, et al. Zipper-interacting protein kinase promotes 29. Sosa MS, Bragado P, Aguirre-Ghiso JA. Mechanisms of disseminated epithelial-mesenchymal transition, invasion and metastasis through cancer cell dormancy: an awakening field. Nat Rev Cancer 2014;14: AKT and NF-kB signaling and is associated with metastasis and poor 611–622. prognosis in gastric cancer patients. Oncotarget 2015;6:8323–8338. 30. Ma M, Chen S, Zhu BY, et al. The clinical significance and risk factors of 52. Ma J, Liu J, Wang Z, et al. NF-kappaB-dependent microRNA-425 solitary lymph node metastasis in gastric cancer. PLoS One 2015;10: upregulation promotes gastric cancer cell growth by targeting PTEN e0114939. upon IL-1beta induction. Mol Cancer 2014;13:40. 31. Deng JY, Liang H. Clinical significance of lymph node metastasis in 53. Burkitt MD, Williams JM, Duckworth CA, et al. Signaling mediated by gastric cancer. World J Gastroenterol 2014;20:3967–3975. the NF-kappaB sub-units NF-kappaB1, NF-kappaB2 and c-Rel differen- 32. Jin X, Zhu Z, Shi Y. Metastasis mechanism and gene/protein expression in tially regulate Helicobacter felis-induced gastric carcinogenesis in gastric cancer with distant organs metastasis. Bull Cancer 2014;101:E1–E12. C57BL/6 mice. Oncogene 2013;32:5563–5573.

1094 Laboratory Investigation | Volume 97 September 2017 | www.laboratoryinvestigation.org